sign in sign up
<<
Home , Andromeda Galaxy, Donatiello I, Dwarf spheroidal galaxy, KKs 3, NGC 404

& Astrophysics manuscript no. aa c ESO 2018 October 12, 2018

Mirach’s Goblin: Discovery of a dwarf spheroidal behind the galaxy David Martínez-Delgado1, Eva K. Grebel1, Behnam Javanmardi2, Walter Boschin3, 4.5, Nicolas Longeard6, Julio A. Carballo-Bello7, Dmitry Makarov8, Michael A. Beasley4, 5, Giuseppe Donatiello9, Martha P. Haynes10, Duncan A. Forbes11, Aaron J. Romanowsky12, 13

1Astronomisches Rechen-Institut, Zentrum für Astronomie der Universität Heidelberg, Mönchhofstr. 12-14, 69120 Heidelberg, Germany 2 School of Astronomy, Institute for Research in Fundamental Sciences (IPM), Tehran, 19395-5531, Iran 3 Fundación G. Galilei - INAF (Telescopio Nazionale Galileo), Rambla J. A. Fernández Pérez 7, E-38712 Breña Baja (La Palma), Spain 4Instituto de Astrofísica de Canarias (IAC), Calle Vía Láctea s/n, E-38205 La Laguna, Tenerife; Spain 5Departamento de Astrofísica, Universidad de La Laguna (ULL), E-38206 La Laguna, Tenerife; Spain 6 Université de Strasbourg, CNRS, Observatoire astronomique de Strasbourg, UMR 7550, F-67000 Strasbourg, France 7Instituto de Astrofísica, Facultad de Física, Pontificia Universidad Católica de Chile, 782-0436 Macul, Santiago, Chile 8 Special Astrophysical Observatory, Nizhniy Arkhyz, Karachai-Cherkessia 369167, Russia 9 Nuovo Orione, 72024 Oria, Italy 10Cornell Center for Astrophysics and Planetary Science, Space Sciences Building, Cornell University, Ithaca, NY 14853 , USA 11 Centre for Astrophysics & Supercomputing, Swinburne University of Technology, Hawthorn VIC 3122, Australia 12 Department of Physics & Astronomy, San José State University, One Washington Square, San Jose, CA 95192, USA 13 University of California Observatories, 1156 High Street, Santa Cruz, CA 95064, USA

ABSTRACT

Context. It is of broad interest for galaxy formation theory to carry out a full inventory of the numbers and properties of dwarf in the Local Volume, both satellites and isolated ones. Aims. Ultra-deep imaging in wide areas of the sky with small amateur telescopes can help to complete the census of these hitherto unknown low galaxies, which cannot be detected by the current resolved stellar population and HI surveys. We report the discovery of , a located one from the Mirach (β And) in a deep image taken with an amateur telescope. Methods. The color– diagram obtained from follow-up observations obtained with the Gran Telescopio Canarias (La Palma, Spain) reveals that this system is beyond the and is mainly composed of old . The absence of young stars and HI emission in the ALFALFA survey are typical of quenched dwarf galaxies. Our photometry suggests a for this galaxy of (m − M) = 27.6 ± 0.2 (3.3 Mpc), although this distance cannot yet be established securely owing to the crowding effects in our color–magnitude diagram. At this distance, Donatiello I’s (MV = −8.3), surface brightness (µV = 26.5 mag arcsec−2) and stellar content are similar to the“classical" companions or . Results. The projected position and distance of Donatiello I are consistent with being a dwarf satellite of the closest S0-type galaxy NGC 404 (“Mirach’s Ghost"). Alternatively, it could be one of the most isolated quenched dwarf galaxies reported so far behind the . Key words. Galaxies:individual:Donatiello I – Galaxies:dwarf – Galaxies:photometry –Galaxies:structure

1. Introduction large scale photometric survey data, such as the Sloan Digi- tal Sky Survey (SDSS; Abazajian et al. 2009), the Panoramic The Λ-CDM paradigm predicts a large number of small dark Survey Telescope and Rapid Response System (Pan-STARRs; arXiv:1810.04741v1 [astro-ph.GA] 10 Oct 2018 matter halos in the Local Volume, but it is unclear how many of Chambers et al. 2017) and more recently the them are associated with luminous baryons in the form of stars. It (DES; Abbott et al. 2018) and the Pan-Andromeda Archaeolog- is thus of broad interest for galaxy formation theory to carry out ical Survey (PAndAs) of the Andromeda galaxy with the wide- a full inventory of the numbers and properties of dwarf galaxies, field imager on the Canada French Hawaii Telescope (CFHT) both satellites and isolated ones. The observations are certainly (McConnachie et al. 2009; Ibata et al. 2014; Martin et al. 2013). biased: the earlier surveys of the Local Group spirals based on In the case of our Galaxy, the main tracers are A-type stars photographic plates were limited in surface brightness to about 2–3 magnitudes fainter than the main sequence (MS)-turnoff of 25.5 mag arcsec−2 (Whiting 2007). For the Milky Way (MW) the old stellar population, which led to the discovery of the ultra- and Andromeda (M31), the recent discoveries of dwarf satellites faint population of dwarf satellites and very diffuse halo stellar were made using stellar density maps of resolved stars, counted sub-structures (Newberg et al. 2002; Willman et al. 2005; Be- in selected areas of the color-magnitude diagrams (CMDs) from lokurov et al. 2006). However, the relatively shallow CMDs from

Article number, page 1 of 11 A&A proofs: manuscript no. aa

The diversity of properties in the local suggests that they have multiple formation paths (Lisker 2009). Some may have formed at early times in the Universe (e.g. Na- gashima et al. 2005; Valcke et al. 2008; Bovill & Ricotti 2009). Others are believed to have been transformed from spiral or ir- regular galaxies by environmental processes, such as tidal inter- actions, harrassment, ram pressure stripping, resonant stripping etc. (Moore et al. 1996, Mayer et al. 2001; Mastropietro et al. 2005; Lisker et al. 2006, 2007; D´ Onghia et al. 2009). An impor- tant way to disentangle environmental from intrinsic processes is to study nearby dwarf galaxies that are isolated from both nearby large neighbors and groups of other dwarfs. The existence of isolated dwarf spheroidal (dSph) galaxies is predicted by some direct hydrodynamical simulations, but the true of these systems remains an open question. The transformation of star-forming, gas-rich dwarf irregular galaxies (dIrr) to gas-poor, passive dSphs is most commonly attributed to the effects of tidal and ram-pressure stripping when a previ- ously isolated dwarf galaxy becomes a satellite of a larger system (e.g. Grebel, Gallagher & Harbeck 2003; Mayer 2010). Dwarfs that form in close proximity to massive halos may furthermore be stripped by local processes and internal heat- ing (e.g. Ocvirk & Aubert, 2011; Nichols & Bland-Hawthorn Fig. 1. Discovery image of Donatiello I dwarf galaxy, obtained with 2011) However, other mechanisms exist that can quench star a 127mm ED doublet refractor f/9 reduced at about f/6 with a Green formation and result in dwarf spheroidal galaxies independent filter and a 2MP cooled CCD camera from the Pollino National Park in of environment. The very shallow potential wells and low virial Southern Italy. The total exposure time is about 6000 seconds obtained temperatures of galaxies in halos with dynamical below 9 by combining (sum and median) images captured on 5,6,7 November ∼ 10 M can lead to an almost complete depletion of the inter- 2010 and 5 October 2013 with the same equipment. The total field of 0 stellar medium due to a combination of feedback and view is this cropped version centered in Do I is 20× 20 . North is up, UV background radiation (e.g., Sawala, Scannapieco & White East is left. 2011; Simpson et al. 2013). As shown by Revaz et al. (2009), such systems can have histories and stellar popu- lations very similar to those of observed dwarf spheroidals near the SDSS (with a g-band limiting magnitude for point sources the MW. A further mechanism that could result in presently iso- of ∼ 21.5) and the significant contamination of the MS-turnoff lated dwarf spheroidal galaxies was recently demonstrated by CMD locus by distant blue galaxies (e.g. see Fig. 1 in Martinez- Benitez-Llambay et al. (2013), who showed that gas can be Delgado et al. 2004), make it difficult to complete the census of stripped from a low- halo via ram-pressure as it is passing faint MW dwarf companions for distances larger than 100 kpc. through the cosmic web at high . Although the M31 photometry from the PAndAS However, such isolated dwarf spheroidals (dSph) are ex- is significantly deeper than the SDSS, its larger distance makes tremely rare in the Local Volume. Geha et al. (2012) found that 9 it prohibitive to resolve the member stars of their companions quenched galaxies with stellar masses of ∼ 10 M and mag- fainter than the level. This means that the satellite nitudes of Mr > −18 (properties that include dEs and dSphs) population (and its stellar streams) can only be traced by means do not exist beyond 1.5 Mpc from the nearest host (∼ 4 virial of the less numerous red-giant branch (RGB) stars. The lack of radii). Indeed, only a few distinct cases of isolated dSph galaxies enough RGB tracers in the CMDs of dwarf galaxies with ab- have been discovered so far: KKR 25 Karachentsev et al. 2001, solute magnitude fainter than MV ∼ −6 makes it very hard to Makarov et al. 2012 at 1.9 Mpc, KKs 3 Karachentsev et al. 2015 detect these lower-mass systems, yielding a cut-off in the lumi- at 2.1 Mpc and Apples 1 Pasquali et al. 2005 at 8.3 Mpc have nosity function of satellites of M31 (see Brasseur et al. 2011; no known galaxies closer than 1 Mpc to them. Karachentseva their Fig. 1). et al. (2010) presented a list of 85 very isolated dwarf galaxy An alternative searching method is the detection of unre- candidates within 40 Mpc, ten of which were classified as dwarf solved (or partially resolved) stellar systems as diffuse light spheroidals. However, these authors pointed out the difficulties structures in low-resolution, deep images taken with small tele- in finding isolated dSphs that cannot be detected in the 21 cm scopes (Mart ınez-Delgado et al. 2016; Besla et al. 2016) in the line surveys and remain undiscovered due to their very low sur- outskirts of the Local Group spirals. Because of the indepen- face brightness. Makarov et al. (2015) carried out spectroscopic dence of the surface brightness on the distance in the nearby uni- observations of three galaxies from the list of isolated dSph verse, this alternative approach also allows the detection of sys- galaxies and concluded that they are not isolated systems. The tems beyond the Local Group, up to distances of 10-15 Mpc (e.g. authors noted that only two (KK 258 and UGC 1703) of the Danieli, van Dokkum & Conroy 2018). Thus, there is still room seven studied so far candidates are isolated dSph galaxies. The for discovery of low surface brightness dwarf galaxies which transition-type dwarf galaxy KK 258 Karachentsev et al. 2014 may lurk in the regions where the detection efficiency drops to resides at 0.8 Mpc from Sdm galaxy NGC 55. According to the very small values (e.g., Koposov et al 2008; see also Whiting Local Volume database Karachentsev et al. 2013 UGC 1703 has et al. 2007 for a detailed discussion of the completeness of the no companions closer than 0.8 Mpc. The number of known very nearby dwarf galaxy population based on diffuse light detection isolated dSph systems does not exceed 5 of more than 1000 of studies). the galaxy population in the Local Volume. The study of such

Article number, page 2 of 11 Martínez-Delgado et al.: Mirach’s Goblin: Discovery of a dwarf spheroidal galaxy behind the Andromeda galaxy

Fig. 2. Left panel: Image of the dwarf galaxy Do I taken with a Takahashi FSQ106 refractor at f/3.6 and a commercial CCD camera QHY90A on 27 November 2016 at the Avalon Merlino Remote Observatory. The total exposure times were 24300 sec in the Luminance filter.The field of view of this cropped image is 123× 123 0 with a pixel scale of 2.89 arcsec pixel−1 (Credits: Matteo Collina & Giuseppe Donatiello). The inner zoomed panel shows the TNG g-band image of the galaxy (see Sec. 2.1). Right panel: r-band image of Do I obtained from GTC observations (see Sec. 2.2). The total field of view is 4.300 × 4.300. North is up, East is left isolated dwarfs populated exclusively by old stars is of great in- include 20x180 sec exposures in the g0-band and 15×180 sec ex- terest to investigate the influence of their environments vs. inter- posures in the r’-band, with a seeing of 0.8"–1" and ∼ 1", respec- nal processes on their star formation histories, and to test models tively. The total field-of-view was 8.6’×8.6’ with a pixel scale of that assume a different origin for these galaxies. 0.252". In this paper, we report the discovery with a small telescope The preprocessing of the data was performed in a stan- of a dwarf spheroidal galaxy projected on a halo field of M31, dard way using IRAF tasks, i.e. dividing the trimmed and bias- which was previously missed by surveys based on resolved star subtracted images by a master flat field produced from multiple counts and on the Hi 21-cm radio line. It was found as part of a twilight sky-flat exposures. For each filter, a final image for pho- new project to search for faint dwarf satellites in the Local Group tometric processing was obtained from a median stack of the with amateur telescopes. preprocessed images. The instrumental magnitudes were derived using daophot 2. OBSERVATIONS AND DATA REDUCTION ii and allstar (Stetson 1987) packages. Sources were detected in the g-band by requiring a 3-σ excess above the local back- Donatiello I (Do I) was first found on 23 September 2016 in a ground, with the list of g-band detections being subsequently visual inspection of a deep amateur image of the Andromeda used for the allstar run on the i0-band image. To obtain clean galaxy region obtained from the composition of a set of im- CMDs and minimize the contamination by faint background ages taken during several years with a refractor ED127mm (with galaxies, sources with the ratio estimator of the pixel-to-pixel an aperture of 12.7-cm at focal ratio f /9) by the amateur as- scatter CHI < 2.0 and sharpness parameters |S | < 1.0 (see Stet- tronomer Giuseppe Donatiello (see Fig.1). The main purpose of son 1987) were kept for further analysis. The instrumental mag- this wide-field mosaic of the area between M31 and M31 was to nitudes were calibrated and transformed to the SDSS system us- detect the stellar streams and M31 satellites previously reported ing stars from the SDSS DR13 catalog. For this calibration we in the PAndAs as diffuse light structures using small amateur used the pipeline developed and used in Javanmardi et al. (2016) telescopes. The detection was confirmed by a visual inspection and also used in Helkel et al. (2017). First, SExtractor (Bertin & of the SDSS DR9 images and follow-up observations using the Arnouts 1996) was run on the image and all the objects in the professional facilities described in Sec. 2.1 and Sec. 2.2. The po- image were detected and their flux was measured. Then the co- sition of the center of this new dwarf galaxy is given in Table 1. ordinates of the detected objects were fed into the SDSS Data Release 13 (Albareti et al. 2016) CrossID and only the stars with ≥ − − 2.1. TNG imaging observations r 15 mag, 0.08 < (r i) < 0.5, and 0.2 < (g r) < 1.4 were se- lected. Then a series of consecutive linear fitting (rsdss vs. rimage We used deep images of the field around the dwarf galaxy ob- and gsdss vs. gimage) with outlier rejection was performed. In each tained with the instrument Device Optimized for the Low Reso- fitting round the stars with ∆ ≡ (magS DS S -magcalibrated) > 2σ lution (DOLoRes) of the 3.58-m Italian Telescopio Nazionale were rejected, where σ is the standard deviation of ∆. The proce- Galileo (TNG)(Roque de Los Muchachos Observatory, La dure was repeated until no 2σ outliers remained. The final stan- Palma, Spain)taken on November 27 2016. These observations dard deviations in ∆ are 0.043 and 0.015 mag for r and g bands,

Article number, page 3 of 11 A&A proofs: manuscript no. aa respectively. They were added in quadrature to the uncertainties of the subsequent measurements.

2.2. GTC imaging observations We took deep images of the Do I dwarf galaxy on January 18 and January 24, 2017 with the Optical System for Imaging and low- Intermediate-Resolution Integrated (OSIRIS)1 in- stalled on the 10.4-m Gran Telescopio Canarias (GTC) (Roque de Los Muchachos Observatory, La Palma, Spain) using an un- vignetted field of view (FOV) of 7.8’×7.8’ and a scale of 0.254" pixel−1. The total exposure time was 22 × 150 sec =3300 sec for each of the three g0,r0, and i0 photometric bands. An 11-point dithering pattern (200 offsets) was used to correct for chip defects and improve image sampling. The seeing was around 1.100 and 0.900 in g0 on January 18, and around 1.100 and 0.9 00 in the g0 and r0-band on January 24, respectively. The resulting r-band image is shown in Figure 2(right panel). We pre-processed the scientific images using IRAF tasks in the standard way (i.e. performing over-scan correction, bias sub- traction and flat field division). In particular, we used the OSIRIS Offline Pipeline Software (Ederoclite & Cepa 2010), a python script that calls IRAF routines in a user-friendly graphical user interface. For each filter, we produced a final image for photo- metric processing by doing a median stack of the preprocessed images. Photometry was obtained with our own pipeline based on the PSF-fitting algorithm of DAOPHOT II/ALLSTAR (Stetson 1987). The final catalog only includes stellar-shaped objects with |sharpness| < 0.5. For the calibration of the images in the g and r0 bands we use the same semi-automatic method introduced in Javanmardi et al. (2016) and briefly described in Sec. 2.1. The final standard deviations in ∆ are 0.031 and 0.045 mag for the r and g bands, respectively, and were added in quadrature to the uncertainties of the subsequent measurements. In order to estimate the completeness of our photomet- ric catalogs, we inserted synthetic stars in the images using DAOPHOT II. The number of artificial stars was 10% of the observed sources for each of the frames. These artificial stars were randomly distributed throughout the chip. The experiment was designed to only include stars with magnitudes in the range 17 < g, r < 28. PSF-fitting photometry was derived for these altered images by applying the same PSF model used for the observed stars. We then measured the fraction of synthetic stars recovered by our procedure and the mean variation of the com- pleteness as a function of the magnitude was derived. Our pho- tometry recovers nearly all the synthetic stars up to g0, r0 ∼ 24 and drops below 80% at g0, r0 ∼ 25.7. We set the limiting mag- Fig. 3. (Upper panel): Photometric error parameter for the resolved stars found DAOPHOT II/ALLSTAR in our GTC images as a function of nitude of our photometry, with a completeness of 50%, at mag = 0 0 magnitude in the g (top) and r-bands (bottom), including only sources 26.7 and 27.0 for the g and r bands, respectively (see Figure 3, with sharpness parameter between -0.5 and 0.5. (Lower panel): Fraction lower panel). of artificial stars recovered as function of input magnitude in g (black line) and r-bands (red line). The horizontal dashed line marks the mag- 3. RESULTS nitude corresponding to a completeness of 50% (see Sec. 2.2) . 3.1. Distance feature of the diagram is the branch (RGB) locus, sug- Figure 4 (lower panel) shows the color–magnitude diagram 0 0 gesting that the stellar population of this dwarf galaxy is domi- (CMD) of Do I for a selected region of 1.8 × 1.8 from the nated by old, metal-poor stars similar to those observed in dSph GTC observations (panel a). The inner region of Do I (within a galaxies in the Local Group. Although we selected a narrow field radius of 0.5 0 centered in the galaxy) is seriously affected by 2 of view centered on the dwarf, there is still contamination by the crowding and was excluded from this diagram. The main brighter RGB stars from the halo of M31 (and some Galactic 1 http://www.gtc.iac.es/instruments/osiris/ 2 See Martínez-Delgado & Aparicio (1997) (e.g. their Fig.1) for a com- and the CMD morphology gradient for the case of an old-population prehensive discussion of the crowding effects in the observational errors dwarf galaxy.

Article number, page 4 of 11 Martínez-Delgado et al.: Mirach’s Goblin: Discovery of a dwarf spheroidal galaxy behind the Andromeda galaxy

Table 1. Properties of the Donatiello I dwarf galaxy

α (J2000) 01h11m40.37s δ (J2000) +34o36003.200 ` 127.65◦ b −28.08◦ (m − M)0 27.6 ± 0.2 Heliocentric Distance 3.3 ± 0.1 Mpc +0.3 MV −8.3−0.3 5.3+0.2 LV 10 −0.3 L −2 µV (mag arcsec ) 26.5 ± 1.0 Ellipticity 0.69 ± 0.05 0 ± rh 0.48 0.15 442±157 pc

foreground stars too) that overlap with this feature in the CMD, as it is also showed in the CMD of the control field situated ∼3.50 North from the center of the galaxy (Fig. 4, panel b). However, the stellar density maps for different magnitude cut-offs show no detection of the dwarf for r-band magnitude r > 24, suggest- ing the bulk of the red stars of this galaxy are fainter than that magnitude. This suggests that this partially resolved system is beyond the Local Group and is not a satellite of the Andromeda galaxy. This also explains its absence in the list of over-density candidates detected with automatic detection algorithms in the PAndAs survey (Martin et al. 2013). A visual inspection of the images from this survey shows Do I was only visible as a par- tially resolved object. The low number of detected upper RGB stars and the M31 halo star contamination makes difficult to estimate the position of the tip of the RGB (TRGB) in the diagram (e.g. by means a Sobel filter; Madore & Freeman 1995). For this reason, we at- tempted to estimate the distance of Do I by the means of CMD fitting following the technique presented in Longeard et al. (in preparation), which models the distribution of stars in the CMD as the sum of a single stellar population and a constant field con- tamination, estimated from the edge of the field of view. Unfortu- nately, both the dwarf galaxy’s CMD and the field contamination model proved to be too noisy to yield sensible results. In particu- lar, we were not able to determine any reliable distance from this analysis, despite testing distance moduli in the 24.0-28.0 range. An attempt at fitting only stars fainter than r0 = 23.5 was made Fig. 4. (Upper panel): Star count map of Do I showing the stars detected in order to discard potential bright stars from the M31 halo that in both g and r-bands in our GTC observations. (Lower panel): De- could affect the fit, but the results were still inconclusive. Nev- reddened (g − r)0 vs. r0 color-magnitude diagram of Do I (panel a) and ertheless, the distribution of stars in the CMD of Do I leaves no its control field (panel b). Panel c) shows a comparison of the Do I CMD ambiguity that Do I stars are present in the field, as also shown with the isochrones (solid lines) with an age of 13 Gyr shifted − in the stellar density map built from our photometry in Figure assuming a distance modulus of (m M)=27.6 mag. The of the isochones are Z=0.0001 (cyan), 0.0003 (red), 0.0005 (blue) and 4 (upper panel). Thus, an accurate distance to Do I cannot be 0.0007 (green) respectively. obtained from the current data with this approach. Since the lack of a well-defined tip of the RGB prevents us from estimating an accurate distance, we explore the possibil- uncertainty) with the distance moduli of NGC 404 reported in ity that Do I is associated with NGC 404, which is at a pro- recent studies ((m − M) = 27.67 ± 0.15; Tikhonov, Galazutdi- 0 jected distance of only 73 (see Sec. 4). For this purpose, we & Aparicio 2003; (m − M) = 27.46 ± 0.02: Lee & Jang obtain a distance modulus for Do I from an eye-ball isochrone 2016). Our isochrone comparison also suggests that a distance fitting of the RGB locus position in its de-reddened CMD, as- larger than ∼ 4 Mpc (i.e. (m − M) > 27.8) is very unlikely for suming this galaxy is mainly composed by old stars. Galactic Do I. extinction values for Do I of Ag=0.206 and Ar=0.143 (Schlafly & Alternatively, there are several brighter red giant star candi- Finkbeiner 2011) were used. Figure 4 (panel c) shows the PAR- dates (in the range 23.7 < r < 24.5)3 which could suggest a SEC isochrones (Bressan et al. 2012) in the range closer distance for Do I. If the actual TRGB lies at the position 0.0001 dex ([Fe/H]=-2.2) to 0.0007 dex ([Fe/H]=−1.3) and an age of 13 Gyr shifted for a distance modulus of (m − M) = 3 Because of the low-quality of our CMD, we cannot confirm with the 27.6 ± 0.2 (dist=3.3 Mpc). This value is consistent (within the current data the presence of brighter, intermediate-age stars above the

Article number, page 5 of 11 A&A proofs: manuscript no. aa

Table 2. The effective V-band surface brightness, µe, and total V-band magnitude measured by GALFIT, the Sérsic index, n, the effective ra- dius, Re, and the axis ratio, b/a. The physical parameters in the lower part of the Table (i.e. effective radius in pc, absolute V-band magnitude, MV , and the ) are calculated by assuming the distance mod- ulus given in Table 1. The V-band quantities were obtained by trans- forming g and r band magnitudes to V magnitudes using Fukugita et al. (1996).

GALFIT SRP −2 µe,V (mag.arcsec ) 26.9 ± 0.07 26.5 ± 1.0 V (mag) 18.25 ± 0.07 - n 0.73 ± 0.02 - Re (arcsec) 20.7 ± 0.3 28.8 ± 9 b/a 0.68 ± 0.01 0.69 ± 0.05 r-band Re (pc) 332 ± 44 442 ± 157 M (mag) −9.3 ± 0.3 −8.3+0.3 V   −0.3 log LV 5.7 ± 0.1 5.3+0.2 10 L −0.3 r- band of the brightest star in this group at r0 ∼ 23.7 (see Figure 4, lower panel), our isochrone comparison would lead to a lower limit for the distance of Do I of 2.2 Mpc. However, since these stars are probably foreground M31 halo stars, we can only assert that Do I is located within a distance range of 2.5 to 3.5 Mpc, and thus a non-member of the Local Group of galaxies.

3.2. Structural properties and surface brightness profile Because of the partially resolved appearance of Do I in our im- GALFIT model ages (see Fig. 5, upper panel), we have used two different ap- proaches to measure its structural properties: i) a GALFIT model fit of the main body of the dwarf as an unresolved, diffuse light system as traced by the TNG data (see Sec. 2.1); ii) the fit of the main body as traced by the resolved stellar population from the GTC stellar photometry. For the first approach, we fit a Sérsic model using the GAL- FIT software Peng et al. 2010. For this purpose, we first cut a sub-image of 400 pixels × 400 pixel of the image centred on the dwarf galaxy, mask the neighbouring foreground stars and background galaxies, and then apply GALFIT. The results are shown in Figure 5 and Table 2. Fig. 5 shows the 400 pixels × 400 pixels r-band image centred on the dwarf galaxy, the Sersic model obtained by GALFIT, and the residual image, which was obtained by subtracting the model from the original image. We note that some resolved stars can be seen in the residual image, Residual which means that they were not taken into account in the GAL- FIT modelling. Although these objects are clearly brighter than the Do I stars detected in the CMD (see Sec. 3.1) and thus are M31 halo star candidates, we cannot reject the possibility that Fig. 5. GALFIT modelling of Do I from the TNG r-band image of Do I. the brightness of the dwarf galaxy was slightly underestimated From top to bottom: the original image centered on the dwarf, the GAL- by GALFIT and the surface brightness profile is also not exact. FIT fitted model, and the residual image obtained by subtracting the Alternatively, to derive the structural parameters of the dwarf model from the original image. galaxy, we use the g and r bands following the technique of Mar- tin et al. (2016) and Longeard et al. (in prep), which accounts for gaps in the spatial coverage of the data. Stars with photometric y0, offsets from a chosen centroid, which are easily converted to uncertainties greater than 0.2 are discarded in both bands. We as- the system’s centroid; rh, the half-light radius (along the major 4 sume an exponential profile for Do I and a flat background con- axis) of the dwarf galaxy; , its ellipticity ; θ, the position angle tamination across the field. The analysis yields posterior proba- of the major axis east to north; and N∗, the number of stars within bility distribution functions (PDFs) for six parameters : x0 and the chosen CMD selection box. The PDFs for rh,  , and θ are

4 RGB locus (e.g. see Martinez-Delgado & Aparicio 1997), which could The ellipticity  is defined here as rh = b (1 − ), with b the minor also bias the distance estimate of Do I. axis of the ellipse.

Article number, page 6 of 11 Martínez-Delgado et al.: Mirach’s Goblin: Discovery of a dwarf spheroidal galaxy behind the Andromeda galaxy

² Θ (deg) 3.0 0.18 0.3 0.15 0.24 0.12 2.5 0.18 l 0.09 o g PDF (

0.12 Σ

0.06 )

( #

0.03 0.06 2.0 s t a r

0.0 0.0 s .

0.5 0.7 0.9 50 100 150 a r

0.15 c m i 0.15 1.5 n 0.12 − 2 ) 0.12 0.09

PDF 0.09 0.06 1.0 0.06

0.03 0.03

0.0 0.0 0.5 0.5 1.0 400 800 0.0 0.5 1.0 1.5 2.0 2.5 3.0 rh (') rh (pc) Radius (')

Fig. 6. Left : Probability distribution functions for some of the structural parameters of Do I from the Resolved Stellar Population (RSP) method: the half light radius rh in angular and physical size, the position angle θ and the ellipticity. Right : Radial density profile within 3 rh of Do I centroid. The number of stars per unit area is represented as a function of the galactocentric radius in arcminutes. The solid curve represents the exponential radial profile drawn from the favoured structural model. shown in Figure 6 and all the structural parameters are summa- lated star. If the completeness is greater than a and b, the star is rized in Table 2. Finally, we work under the assumption that the flagged. A total number of N∗, j are accumulated in this way. Fi- Do I half-light radius should not be smaller than 0.1 arc minutes nally, the fluxes of all simulated stars, flagged or not, are summed on the sky. to give the total luminosity of the system. This procedure is re- The radial density profile of the dwarf galaxy is displayed in peated several times in order to obtain a PDF in luminosity and Figure 6 and compared to the star counts in elliptical annuli. The absolute magnitude. theoretical exponential profile drawn from the favoured struc- tural parameters is also shown in Figure 6, in red, and shows good agreement with the binned data. Our method to estimate the luminosity follows the analysis described in Martin et al. (2016) and can be interpreted as the Table 2 shows a comparison of the results for the structural statistical determination of the total luminosity of any system parameters obtained with these two different methods. The up- with the same structural properties as Do I. At each iteration, a per part of Table 1 gives the effective surface brightness, µe, to- number of stars N∗, j is randomly drawn following the N∗ poste- tal , the Sersic index, n, the effective radius, rior distribution function obtained through our structural fitting Re, and the axis ratio, b/a. The physical parameters in the lower procedure. The distance modulus is also randomly drawn from a part of Table 2 (i.e. effective radius in pc, absolute magnitude, Gaussian distribution of mean (m− M) = 27.6 and standard devi- MV , and the luminosity) are calculated by assuming the distance ation δ(m−M) = 0.2. We also choose a stellar population of [Fe/H] modulus ((m − M)0 = 27.6 ± 0.2 mag) for the dwarf galaxy. = −2.0, 12 Gyr and solar alpha abundance, which corresponds to The resulting absolute magnitudes, surface brightness and effec- the typical population one should expect from a MW/M31 dSph tive radii are also plotted in Figure 7, in comparison with those with Do I’s total luminosity (Kirby et al. 2013). The choice of a for known Milky Way and M31 dwarf galaxy companions. In- slightly different age, metallicity or [α/Fe] would not impact the terestingly, the structural parameters obtained from the resolved inference of the luminosity of the system. A star is then simu- stars and diffuse light component yield similar results, suggest- lated along this isochrone : its g0 and r0 magnitudes are checked ing that Do I has an absolute magnitude, surface brightness, and : first, they should fall within the magnitude and color limits of size similar to some of the “classical" dSph companions of the the data (i.e. the min and max g0 − r0 and g0 in our CMD). For Milky Way (e.g. Draco or Ursa Minor). Both methods provide simulating the completeness in the CMD, two random numbers the same value for its ellipticity, confirming this galaxy is very a and b between 0 and 1 are randomly drawn and respectively elongated as previously noted in the stellar density map plotted compared with the completeness at the g0 and r0 of the simu- in Fig. 4 (upper panel).

Article number, page 7 of 11 A&A proofs: manuscript no. aa

Derivation of an upper limit to the HI mass for an individual LG dwarf galaxies object, for example as is described by Spekkens et al. (2014), Donatiello I. (GALFIT) Donatiello I. (RSP) requires assumptions about the source extent and spectral defini-

3 tion. Given that the g-band half-light radius is 24", we expect the 10 H i emission to be significantly smaller than the effective ALFA beam (4 0) so that the H i would be unresolved. For comparison ]

c with Spekkens et al.(2014), we smooth the ALFALFA data to p

[ −1 15 km s in velocity and examine the ALFALFA cube over an e R ellipse following the standard ALFALFA minimum region of 7 0 0 102 × 7 (Haynes et al. 2018 in prep.) over ranges of relevant veloc- ities. The complication is that the noise increases in the presence of Galactic H i line emission; the standard ALFALFA drift scan strategy is not optimal for mapping the Galactic H i emission al- though moderately strong, (nearly) unresolved features are iden- 3 4 5 6 7 8 9 tified (Giovanelli et al 2013; Adams et al. 2013). Analysis of the log10(LV/L ) ALFALFA grid at the position of Do I gives an r.m.s. noise per beam solid angle of 2.0 mJy at 15 km s−1 resolution. For a sig-

LG galaxies nal of that width, the 5-σ upper limit of the integrated line flux −1 −1 18 Donatiello I. (GALFIT) density at 15 km s is 0.15 Jy km s , translating to an upper Donatiello I. (RSP) 4 2 limit on the H i mass of 3.5 ×10 D M where D is the distance 5 20 in Mpc. For a distance of 3.3 Mpc, this translates to 3.8 × 10 ]

2 M . c e

s 22 It should be emphasized that this limit assumes that the HI c r

a emission in Do I is unresolved both spatially and spectrally by /

g 24 the ALFALFA survey, both of which are reasonable assump- a m

[ tions.

e 26

28 4. DISCUSSION

30 We have reported the discovery of Do I, a faint stellar system 2 3 at a projected distance of one degree from the Mirach (β And) 10 10 star. Our data suggest a distance range of 2.5–3.5 Mpc for Do I, Re [pc] and its structural parameters and absence of HI are consistent 20 with those of a dwarf spheroidal galaxy, similar to the known LG dwarf galaxies Donatiello I. (GALFIT) companions of the Milky Way or M31. Donatiello I. (RSP) Do I lies 72.4 arcmin on the sky from NGC 404, which corre- 22 sponds to a projected separation of 65 kpc if we adopt the TRGB

] distance of NGC 404 equal to 3.1 Mpc (Lee & Jang 2016). In that 2 c

e case, the absolute V-band magnitude of Do I is −8.2 mag, which

s 24 c

r is about 9.6 mag fainter than NGC 404 itself, MV = −17.79. a /

g Interestingly, NGC 404 has likely experienced a recent interac- a tion/merger, leading to a kinematically decoupled core and an m 26 [

e HI shell around the galaxy (del Rio et al. 2004; Bouchard et al. 2010). Thilker et al. (2010) showed ongoing star formation in an 28 extended disk-like HI “ring" around this visi- ble in the GALEX images. del Rio et al. (2004) argued that this gaseous ring around NGC 404 is the remnant of a merger with a 30 2.5 5.0 7.5 10.0 12.5 15.0 17.5 dwarf which took place some 900 Myr ago. The MV [mag] proximity of Do I and its elongated shape lead one to speculate that it might have been involved in that interaction or potentially Fig. 7. A comparison between the properties of Do I and the known have been affected by it. A more accurate distance and a radial dwarf galaxies of the Local Group (McConnachie 2012). Top: effective velocity measurement are needed to shed more light on this in- radius in pc vs. the logarithm of V band luminosity in L . Middle: V teraction scenario. band surface brightness in mag arcsec−2 vs. effective radius in pc. Bot- tom: V band surface brightness in mag arcsec−2 vs. absolute V band The existence of such systems of dwarfs is not unusual. Tully magnitude. The V band quantities were obtained by transformations et al. (2006) pointed out that within 3 Mpc, except for KKR 25 given in Fukugita et al. (1996), and the effective radius is the average of and as we know KKs 3, every dwarf is associated with either a the g and r band measurements. luminous group or an association of dwarfs. Makarov & Uklein (2012) compiled a list of systems where the luminosity of both components are lower than the luminosity of the Small Magel- 3.3. H i emission lanic Cloud. Such systems are quite numerous in the Local Su- percluster and are located in low-density regions. In recent years, Do I was not detected by the ALFALFA survey (Haynes et al., the number of discoveries of multiple dwarf systems have been in preparation) but we have examined the ALFALFA grid at made. For instance, Makarova et al. (2018) discovered a satel- the relevant location to derive an upper limit to the H i flux. lite of an isolated irregular dwarf at a distance of 9.2 Mpc. This

Article number, page 8 of 11 Martínez-Delgado et al.: Mirach’s Goblin: Discovery of a dwarf spheroidal galaxy behind the Andromeda galaxy

Luminosity, mag Morphology -21 -20 -19 -18 -17 -16 -5 0 5 10

2.5 2.5

2 2

1.5 1.5 KKR25 KKR25 1 1

0.5 0.5

0 0 , Mpc MW MW , Mpc

SG M 31 M 31 SG Y And XVIII Y -0.5 And XVIII -0.5 Tucana Cetus Cetus -1 -1 KKs 3 KKs 3 -1.5 Do I Do I -1.5

-2 -2

-2.5 -2.5 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 22.5 0.511.5 -1-0.50 -2.5-2-1.5 X , Mpc Z , Mpc SG SG

Fig. 8. Super-galactic coordinate map of the distribution of galaxies within a 5 Mpc cube roughly centered on the Local Group. The size of filled circles is scaled to the luminosity of a galaxy. The T-values morphological classification of each system is color-coded according to the color-bar scale. The big circle encompassing the Local Group corresponds to the zero-velocity surface Karachentsev et al. 2009, which separates the collapsing region of space around the Local Group from the cosmological expansion. Known highly isolated dSph systems are marked with a red star as well as Do I. The short line from Do I indicates the distance uncertainty from 2.5–3.5 Mpc. In this map, the position of NGC 404 approximately overlaps with those of Do I at 3.5 Mpc. pair is characterized by a projected separation of 3.9 kpc and ies surrounding the Local Group is given in Table 3. All the absolute V-magnitude of −13.3 and −9.4 mag for the main com- isolated dSphs within 2 Mpc of the MW have precise distance ponent and the satellite, respectively. Carlin et al. (2016) found a measurements, which makes it possible to assert their isolation satellite with Mg = −7.4 at a projected distance of ∼ 35 kpc from with great confidence (see column 7 in Table 3). The dSph galax- dwarf spiral NGC 2403 in the framework of the MADCASH sur- ies Cetus and And XVIII lie inside the zero-velocity surface of vey. This galaxy has an old metal-poor stellar population similar the Local Group, while Tucana is located just beyond the Local to ultra-faint galaxies of the Local Group. Chengalur & Pustilnik Group boundary at a distance of 0.91 Mpc from Milky Way and (2013) discovered a triplet of extremely gas-rich galaxies near 1.38 Mpc from Andromeda (Jacobs et al. 2009). However, due the center of the Lynx-Cancer . The authors argued that the to the uncertain nature of their orbits, past interactions for these appearance of such objects in low-density regions is a conse- cannot be excluded. Teyssier et al. (2012) have analyzed the or- quence of the slow evolution of the structure formation in com- bits of potential dwarf galaxy halos in the Via Lactea II simu- parison with the denser regions. lation to determine the likelihood that presently isolated dwarf The large uncertainty of its distance (see Sec. 3.1) and the galaxies had past interactions with the Milky Way or M31. They lack of a measurement prevent a clear association found that some 13% of dwarf galaxies with present distances of Do I with NGC 404. Another possibility is that Do I could between 300 and 1500 kpc could be classified as escapees, mak- be an isolated dSph galaxy beyond the Local Group. Geha et ing past interactions likely. However, at a present distance larger al.(2002) found that quenched dwarf galaxies are always found than 2.5 Mpc, the origin of Do I as a past satellite of the Milky within 1.5 Mpc of a more massive host. However, while most of Way or M31 appears very unlikely, which would make it the first the known classical dSph galaxies have been discovered within unambiguous example of an isolated dwarf spheroidal galaxy a distance of 300 kpc from the Milky Way or M31 (where envi- (if not a satellite of NGC 404). Apart from the gas fraction ronmental effects are important), there are already several known and the resultant star formation history, which distinguish dwarf examples of presently isolated dwarf spheroidal galaxies. In the spheroidal and dwarf irregular galaxies (Grebel et al. 2003), past case of a distance for Do I of 2.5 Mpc, it would be one of the tidal interactions can also be imprinted on a galaxy’s stellar kine- most isolated dSph galaxies reported so far. matics (e.g., Muñoz, Majewski & Johnston 2008). While not all dwarf spheroidals show such tidal features (Peñarrubia et al. Fig. 8 shows a map in the super-galactic coordinates of the 2009), their presence in Do I could be an important further clue distribution of galaxies within a 5 Mpc cube roughly centered to its origin as a truly isolated dwarf spheroidal, or as a system on the Local Group. Known highly isolated dSph systems are with past interactions of some kind. marked with red stars, including Do I. The short line marked from Do I indicates its position for the distance range from Because of its distance range and its crowding level, deeper, 2.5–3.5 Mpc discussed in Sec. 3.1. The list of isolated galax- high resolution imaging and radial ve-

Article number, page 9 of 11 A&A proofs: manuscript no. aa

Table 3. Isolated dwarf galaxies around the Local Group

Main Disturber Closest neighbor Name MB MV Distance Ref. ∆ MB ∆ (mag) (mag) (Mpc) (Mpc) (mag) (Mpc) KKs 3 −10.8 −12.3 2.12 ± 0.07 1 MW 2.12 IC 3104 −14.9 1.02 KKR 25 −9.4 −10.7 1.92 ± 0.06 2 M 31 1.86 KK 230 −9.6 0.98 Tucana −9.2 −9.7 0.92 ± 0.02 3 MW 0.91 ESO 245–007 −9.5 0.60 Cetus −10.1 −10.7 0.79 ± 0.04 3 M 31 0.69 IC 1613 −14.4 0.22 And XVIII −10.4 1.33 ± 0.08 4 M 31 0.55 And XIX 0.44 References: (1)Karachentsev et al. 2015, (2)Makarov et al. 2012 , (3)Jacobs et al.(2009), (4) Makarova et al.(2017) locity observations are needed to constrain the distance of Do I Carlin, J. L., Sand, D. J., Price, P., et al. 2016, ApJ, 828, L5 and definitively determine if it is a NGC 404 companion. The Brasseur, C. M., Martin, N. F., Macció, A. V., Rix, H. W., Kang, X. 2011, ApJ, 743, 179 discovery of Do I with an amateur telescope also shows that Bovill, M. S., Ricotti, M., 2009,ApJ, 693, 1859 ultra-deep imaging in wide sky areas is still very valuable for de- Chengalur, J. N., & Pustilnik, S. A. 2013, MNRAS, 428, 1579 tecting faint, hitherto unknown low surface brightness galaxies Danieli, S., van Dokkum, P., & Conroy, C. 2017, arXiv:1711.00860 that were not detected by means of resolved stellar populations del Río, M. S., Brinks, E., & Cepa, J. 2004, AJ, 128, 89 i DÓnghia, E., Besla, G., Cox, T. J., Hernquist, L., 2009, Nature, 460, 605 or H surveys. In fact, the wide fields and depths (2–3 magnitudes Dooley, G. A., Peter, A. H. G., Carlin, J. L., et al. 2017, MNRAS, 472, 1060 deeper than the POSS-II survey) of these small telescopes make Ederoclite, A., & Cepa, J. 2010, Astronomical Data Analysis Software and Sys- them ideal for uncovering galaxies within the Local Group and tems XIX, 434, 253 beyond (∼ 10 Mpc) at low surface brightness levels, including Fukugita, M., Ichikawa, T., Gunn, J. E., et al. 1996, AJ, 111, 1748 Gavazzi, G., Fumagalli, M., Cucciati, O., Boselli, A. 2010, A&A, 517, A73 in the outskirts of the Local Group where Λ-CDM simulations Geha, M., Blanton, M. R., Yan, R., Tinker, J. L. 2012, ApJ, 757, 85 predict a significant population of free-floating, still undetected, Giovanelli, R., Haynes, M. P., Adams, E. A. K., et al. 2013, AJ, 146, 15 isolated dwarf galaxies. These systems are of great interest for Grebel, E. K., Gallagher, J. S., III, & Harbeck, D. 2003, AJ, 125, 1926 modern cosmological scenarios of galaxy formation, and could Henkel, C., Javanmardi, B., Martínez-Delgado, D., Kroupa, P., & Teuwen, K. 2017, A&A, 603, A18 still remain undetected in large scale optical and radio surveys Jacobs, B. A., Rizzi, L., Tully, R. B., et al. 2009, AJ, 138, 332 due to their extremely faint surface brightness and low gas con- Javanmardi, B., Martinez-Delgado, D., Kroupa, P., et al. 2016, A&A, 588, A89 tent. Jacobs, B. A., Rizzi, L., Tully, R. B., et al. 2009, AJ, 138, 332 Karachentsev, I. D., Sharina, M. E., Dolphin, A. E., et al. 2001, A&A, 379, 407 Acknowledgements. We acknowledge M. Cecconi and D. Carosati for their sup- Karachentsev, I. D., Kashibadze, O. G., Makarov, D. I., & Tully, R. B. 2009, port during the follow-up of Do I with the TNG. We also thank N. Martin, MNRAS, 393, 1265 C. Wittmann and G. Morales. DMD and EKG acknowledge support by Son- Karachentsev, I. D., Makarov, D. I., & Kaisina, E. I. 2013, AJ, 145, 101 derforschungsbereich (SFB) 881 “The Milky Way System” of the German Re- Karachentsev, I. D., Makarova, L. N., Tully, R. B., Wu, P.-F., & Kniazev, A. Y. search Foundation (DFG), particularly through sub-project A2. TSC acknowl- 2014, MNRAS, 443, 1281 edges the support of a National Science Foundation Graduate Research Fellow- Karachentsev, I. D., Makarova, L. N., Makarov, D. I., Tully, R. B., & Rizzi, L. ship. JAC-B acknowledges financial support from CONICYT-Chile FONDE- 2015, MNRAS, 447, L85 CYT Postdoctoral Fellowship 3160502. MAB acknowledges support from grant Karachentseva, V. E., Karachentsev, I. D., & Sharina, M. E. 2010, Astrophysics, AYA2016-77237-C3-1-P from the Spanish Ministry of Economy and Competi- 53, 462 tiveness (MINECO) and from the Severo Ochoa Excellence scheme (SEV-2015- Kirby, E. N., Cohen, J. G., Guhathakurta, P., et al. 2013, ApJ, 779, 102 0548). MPH is supported by grants from the US NSF/AST-1714828 and the Ibata, R. et al. 2014, ApJ, 780, 128 Brinson Foundation. GD thanks Matteo Collina and Tim Stone for the use of Koposov, S., Belokurov, V., Evans, N. W., Hewett, P. C., Irwin, M. J., Gilmore, their remote observatory for this work. AJR was supported by National Sci- G., Zucker, D. B., Rix, H.-W., Fellhauer, M., Bell, E. F., Glushkova, E. V.. ence Foundation grant AST-1616710 and as a Research Corporation for Science 2008, ApJ, 686, 279 Advancement Cottrell Scholar. We acknowledge the usage of the HyperLeda Lee, M. G., & Jang, I. S. 2016, ApJ, 822, 70 database (http://leda.univ-lyon1.fr). Based on observations made with the Italian Lisker, T. 2009, AN, 330, 1403 Telescopio Nazionale Galileo (TNG), operated by the Fundación Galileo Galilei Lisker, T., Glatt, K., Westera, P., Grebel, E. K. 2006, AJ, 132, 2432 of the INAF (Istituto Nazionale di Astrofisica), and with the Gran Telescopio Ca- Lisker, T., Grebel, E. K., Binggeli, B., Glatt, K. 2007, ApJ, 660, 1186 narias (GTC) under Director’s Discretionary Time. Both telescopes are installed Madore, B. F., & Freedman, W. L. 1995, AJ, 109, 1645 in the Spanish Observatorio del Roque de los Muchachos (La Palma, Spain) of Makarov, D. I., & Uklein, R. I. 2012, Astrophysical Bulletin, 67, 135 the Instituto de Astrofísica de Canarias. The authors also acknowledge the use Makarov, D., Makarova, L., Sharina, M., et al. 2012, MNRAS, 425, 709 of IRAF, the Image Reduction and Analysis Facility, a general purpose software Makarov, D. I., Sharina, M. E., Karachentseva, V. E., & Karachentsev, I. D. 2015, system for the reduction and analysis of astronomical data. IRAF is written and A&A, 581, A82 supported by the National Optical Astronomy Observatories (NOAO) in Tuc- Makarov, D., Prugniel, P., Terekhova, N., Courtois, H., & Vauglin, I. 2014, A&A, son, Arizona. NOAO is operated by the Association of Universities for Research 570, A13 in Astronomy (AURA), Inc. under cooperative agreement with the National Sci- Makarova, L. N., Makarov, D. I., Karachentsev, I. D., Tully, R. B., & Rizzi, L. ence Foundation. Funding for SDSS-III has been provided by the Alfred P. Sloan 2017, MNRAS, 464, 2281 Foundation, the Participating Institutions, the National Science Foundation, and Makarova, L. N., Makarov, D. I., Antipova, A. V., Karachentsev, I. D., & Tully, the U.S. Department of Energy Office of Science. R. B. 2018, MNRAS, 474, 3221 Mastropietro, C., Moore, B., Mayer, L., Wadsley, J., Stadel, J. 2005, MNRAS, 363, 509 Martin, N. F., Ibata, R. A., McConnachie, A. W., Dougal, M., A., Ferguson, A. N.; Irwin, M. J.; Lewis, G. F.; Fardal, M. A. 2013, ApJ, 776, 80 References Martin, N. F., Ibata, R. A., Lewis, G. F., et al. 2016, ApJ, 833, 167 Martínez-Delgado, D. & Aparicio, A. 1997, ApJ, 480, L107 Abazajian, K. N., Adelman-McCarthy, J. K., Agüeros, M. A., et al. 2009, ApJS, Martínez-Delgado, D., Gómez-Flechoso, M. A., Aparicio, A.; Carrera, R. 2004, 182, 543-558 ApJ, 601, 242 Albareti, F. D., Allende Prieto, C., Almeida, A., et al. 2017, ApJS, 233, 25 Martínez-Delgado, D., Läsker, R., Sharina, M., et al. 2016, AJ, 151, 96 Belokurov, V. et al. 2006, ApJ, 647, 111 Mayer, L., Governato, F., Colpi, M., Moore, B., Quinn, T., Wadsley, J., Stadel, Benítez-Llambay, A., Navarro, J. F., Abadi, M. G., et al. 2013, ApJ, 763, L41 J., Lake, G., 2001, ApJ, 559, 754 Bertin, E., & Arnouts, S. 1996, A&AS, 117, 393 Mayer, L. 2010, Advances in Astronomy, 2010, 278434 Besla, G., Martínez-Delgado, D., van der Marel, R. P., et al. 2016, ApJ, 825, 20 McConnachie, A.W. et al. 2009, Nature, 461, 66 Bressan, A., Marigo, P., Girardi, L., et al. 2012, MNRAS, 427, 127 McConnachie, A.W. 2012, AJ, 144, 4 Bouchard, A., Prugniel, P., Koleva, M., & Sharina, M. 2010, A&A, 513, A54 Moore, B., Katz, N., Lake, G., Dressler, A., Oemler, A., 1996, Nature, 379, 613

Article number, page 10 of 11 Martínez-Delgado et al.: Mirach’s Goblin: Discovery of a dwarf spheroidal galaxy behind the Andromeda galaxy

Muñoz, R., Majewski, S. R., Johnston, K. V. 2008, ApJ, 679, 346 Nagashima, M., Yahagi, H., Enoki, M., Yoshii, Y., Gouda, N. 2005, ApJ, 634, 26 Newberg, H. J. et al.. 2002, ApJ, 569, 245 Nichols, M., Bland-Hawthorn, J. 2011, ApJ, 732, 17 Ocvirk, P., Aubert, D. 2011, MNRAS, 417, 93 Pasquali, A., Larsen, S., Ferreras, I., et al. 2005, AJ, 129, 148 Peng, C. Y., Ho, L. C., Impey, C. D., & Rix, H.-W. 2010, AJ, 139, 2097 Peñarrubia, J., Navarro, J. F., McConnachie, A. W., & Martin, N. F. 2009, ApJ, 698, 222 Revaz, Y., Jablonka, P., Sawala, T., et al. 2009, A&A, 501, 189 Sawala, T., Scannapieco, C., & White, S. 2012, MNRAS, 420, 1714 Schlafly, E. F., & Finkbeiner, D. P. 2011, ApJ, 737, 103 Simpson, C. M., Bryan, G. L., Johnston, K. V., et al. 2013, MNRAS, 432, 1989 Spekkens, K., Urbancic, N., Mason, B. S., Willman, B., & Aguirre, J. E. 2014, ApJ, 795, L5 Stetson, P.B. 1987, PASP, 99, 191 Teyssier, M., Johnston, K. V., & Kuhlen, M. 2012, MNRAS, 426, 1808 Thilker, D. A., Bianchi, L., Schiminovich, D., et al. 2010, ApJ, 714, L171 Tikhonov, N. A., Galazutdinova, O. A., & Aparicio, A. 2003, A&A, 401, 863 Tully, R. B., Rizzi, L., Dolphin, A. E., et al. 2006, AJ, 132, 729 Valcke, S., de Rijcke, S., Dejonghe, H., 2008,MNRAS, 389, 1111 Wheeler, C., Oñorbe, J., Bullock, J. S., et al. 2015, MNRAS, 453, 1305 Whiting A.B., Hau, G. K. T.; Irwin, M., Verdugo, M. 2007, AJ, 133, 715 Willman, B., Dalcanton, J. J., Martinez-Delgado, D., et al. 2005, ApJ, 626, L85

Article number, page 11 of 11